390-05 Election Analysis 1996
~
\
~
~
THE MIAMI BEACH MAYOR AND COMMISSIONERS
ELECTION OF NOVEMBER 1995:
AN ETHNIC ANALYSIS
by
>
Abraham D. Lavender, Ph.D.
and
Chris Girard, Ph.D.
.'
Department of Sociology and Anthropology
Florida International University
North Miami, Florida 33181
January 1996
,
i
Florida International University
To:
Miami Beach Mayor Seymour Gelber
Commissioners
Candidates for Mayor and Commission, November 1995
City Manager
City Clerk
City Attorney
~IDAD )~
~LlA-
Dr. Abe Lavender
Department of Sociology and Anthropology
Florida International University
From:
Re:
November 1995 Elections
Date:
February 12, 1996
Enclosed is an analysis of the elections for Mayor and
Commissioners in Miami Beach in November 1995. The vote on renaming
Ocean Beach Park also is included. The report hopefully can help us
better understand the demographic and political changes tp.king
place in our city. The analysis is based on the sociological and
statistical methods routinely used in the analysis of elections--
bivariate ecological regression analysis and homogeneous precinct
analysis (where available). The methodology is similar to that used
in my 1994 study which analyzed elections in Miami Beach from 1983
through 1993.
The 1994 study led to some rather negative personal reactions
against me by a few people who did not agree with the results and
their explanations. Despite those reactions, as an academician I
continue to believe that knowledge has the potential to help us
solve areas of conflict. I hope that this information will be
useful in helping us reach areas of agreement and avoiding areas of
divisiveness in our community. I beseech you to use the information
in a positive manner.
Nonh Miami Campus. Nonh Miami, Florida 33181
E.paI Opponun;tyIEquoI ""'" Emp.".. ond h........
f
THE MIAMI BEACH MAYOR AND COMMISSIONERS
ELECTION OF NOVEMBER 1995: AN ETHNIC ANALYSIS
by
Abraham D. Lavender, Ph.D.
and Chris Girard, Ph.D.
Department of Sociology and Anthropology
Florida International University
January 1996
On November 7, 1995, the City of Miami Beach held an election
for Mayor and two Commissioners. The mayor serves a two-year term,
with elections being held in November of every odd-numbered year.
Six commissioners serve four-year terms, but the terms are
staggered so that three commission seats are up for election every
two years at the same time as the election for mayor.
In November 7, 1995, there were two candidates for mayor. The
incumbent mayor, retired judge Seymour Gelber, ran for reelection
after serving two consecutive terms from November 1991 to November
1995. The other candidate was Andrew Delaplaine, the editor of a
local small newspaper, with no previous elective experience in
Miami Beach politics.
The three commission seats up for election in November 1995
were the seats held by Commissioners David Pearlson, Martin
Shapiro, and Susan Gottlieb. All three commissioners ran for
reelection. Commissioner David Pearlson, who had served a four-year
term from November 1991 to November 1995, had no opponent, and
hence received another four-year term without an election.
Commissioner Martin Shapiro was a six-year incumbent, having first
been elected in 1989 for a two-year term, and being reelected in
1991 for a four-year term. He was opposed by Matti Bower, a local
community activist with experience in civic and educational
activities. Commissioner Susan Gottlieb was first elected in 1991,
for a four-year term, and was opposed by Ada Llerandi, a local
hotelier who served on one of the city's most important committees.
The November 1995 election was the first election held after
federal judge James Kehoe ruled, earlier in 1995, against a group
of Hispanic plaintiffs who argued that the city's at-large
elections kept Hispanics from winning because of ethnically
polarized voting. A previous study by these authors indicated that
ethnically polarized voting had been a common pattern in Miami
Beach elections since the city had developed- a sizable Hispanic
voting population in the last decade or so (Lavender and Girard,
September 1994) .
In the election of November 1995, none of the incumbents, but
two of the challengers, were Hispanic--Ada Llerandi who challenged
incumbent Commissioner Susan Gottlieb, and Matti Bower who
challenged incumbent Commissioner Martin Shapiro. Hence, this study
uses the same two techniques that were used in the previous study--
homogeneous precinct analysis and bivariate ecological regression
analysis--to determine ethnic voting patterns for these two
elections in November 1995. In addition, this study also analyses
the other two issues, the mayor's race and the question of whether
to change the name of Ocean Beach Park to Marjorie Stoneman Douglas
Park, in order to determine whether these races exhibited ethnic
voting patterns.
Voters List
The analyses herein are based on a listing of all individual
voters in Miami Beach, coded for whether or not they voted in
November 1995. Hence, these analyses are not based simply on group
(precinct) level results. One slight disadvantage of using this
individual data is that it takes the elections board several weeks
to post on the voting roll whether or not each person voted, and by
that time a few other changes have taken place--including a few
people who voted in November being removed from the voting list.
[The voting list is not "frozen" on the day of the election, but is
changed almost daily as people or added to or removed from the
voting list]. However, the number of voters "lost" from the
analyses was only 35, from a listing of 34,812 to 34,777. This is
a loss of only one-tenth of one percent. The "percentage of votes
counted" thus might be off by about one-tenth of one percent or
less. On the other hand, one advantage of using the individual
listing is that the 759 voters who voted by absentee ballots are
included in these analyses, and are counted in the precincts in
which they live. If the original precinct data had been used, these
759 voters would be lost from the analyses because it would not be
possible to put them into the correct precinct or to determine
whether they were Hispanics or non-Hispanics. The loss of voters is
very small (35), and the gain (759) is much greater than the loss
because we are able to use individual instead of group data in
determining ethnic status. In addition, and perhaps more
importantly, the individual data allows us to obtain results for
ethnic subgroups, age subgroups, and a number of other demographic
subgroups, results that are not possible from precinct data.
Definitions
Before January 1995, voters gave their place of birth when
they registered as voters. Dade County election officials then gave
Hispanic [Latin] registration figures based on foreign-born voters
only, those voters born in a Spanish-speaking country or Puerto
Rico. Voters not born in a Spanish-speaking country or Puerto Rico
were counted as "non-Latin," regardless of their Spanish heritage
or identity. In our September 1994 study, we estimated the number
of United States-born Hispanic voters (voters born outside a
Spanish-speaking country, about 95% of whom were born in the United
States) by using a Distinctive Hispanic Names Technique. These two
groups are combined to estimate the total number of Hispanic
voters.
2
Beginning in January 1995, the voter registration process no
longer asked place of birth, but asked voters to indicate their
ethnic identity. Hence, the total number of Hispanic voters for the
November 1995 election consists of a combination of foreign-born
and estimated United States-born (using the Distinctive Hispanic
Names Technique) for those registered prior to January 1995, plus
those who registered during 1995 and identified their ethnicity as
Hispanic. The total of 12,945 Hispanic voters consists of 9,626
foreign-born, 2,077 estimated United States-born, and 1,242 self-
identified as Hispanic (first registered in 1995). In order to
compare voter turnouts for different ethnic groups in Miami Beach,
a Distinctive Jewish Names Technique is used to obtain a sample of
Jewish voters, and a list of Distinctive Anglo Names is used to
obtain a sample of Anglo (non-Hispanic, non-Jewish, non-black)
voters. These two listings give samples, although the Distinctive
Jewish Names Technique has been used frequently for fifty years to
predict Jewish population sizes.
Estimated Ethnic Voter Registration Numbers
Using the techniques described above, the following are
estimates of the ethnic breakdown of voters by percentages for the
last few years, including the November 7th, 1995, election
(Lavender, 1991: 11; Lavender, 1993, 10):
Feb. 1991 Jan. 1993 Nov. 1995
Total Number of Voters 37,014 38,552 34,777*
Hispanic Number 10,334 11,954 12,945
Non-Hispanic Number 26,680 26,598 21,832
Jewish Number 19,687 17,900 13 , 142
"Other" Number** 7,797 9,457 9,290
Percent Hispanic 27.9 31.0 37.2
Percent Non-Hispanic 72.1 69.0 62.8
Estimated Percent Jewish 53.2 46.4 37.8
Estimated Percent "Other" 21.1 24.5 26.7
*Remember that the actual number was 34,812, thirty-five more.
** "Other" means non-Hispanic and non-Jewish. The total is about
101.7% because about 600 voters are both Hispanic and Jewish, are
counted in both groups, and hence lead to an overlap.
Method of Analysis
As in the earlier (September 1994) study, two methods of
analysis are used: Ecological Regression Analysis and Homogeneous
3
Precinct Analysis. The Ecological Regression Analysis is based on
precinct analyses, comparing the percentage of Hispanic voters in
each precinct to the percentage of votes received by Hispanic
candidates in each precinct. A complex statistical procedure
predicts the "best fit" between these two variables. Precincts are
weighted for size so that each precinct influences the statistical
procedure according to its size.
Homogeneous Precinct Analysis takes precincts which are
heavily of one ethnic group, and looks at the votes for each
candidate in these "ethnic" precincts. In Miami Beach, there are a
cluster of precincts which are heavily non-Hispanic or "Anglo"
(slightly over 80%), and hence these precincts can be analyzed.
There are precincts which are majority Hispanic, and there are more
precincts in which the majority of votes cast were Hispanic, but
there are no precincts that are predominantly Hispanic. Hence, we
cannot analyze "Most Hispanic" precincts. The "Most-Hispanic"
results are presented on the following pages, along with the
analyses of most-"Anglo" precincts, for information purposes only,
Double checking ecological regression analysis and homogeneous
precinct analysis against each other gives a good idea of whether
the two techniques are close estimates of the actual voting
resul ts. In the middle of each "Homogeneous Precinct Analysis"
page, these two techniques are combined. As is shown, in each of
these four elections the two sets of analysis were very close to
each other in their estimates, suggesting that the analyses are
very close estimates of reality. More details on these two methods
can be found in the September 1994 report.
Voter Turnout Rates
Voting results depend not only on absolute numbers of
registered votes, but on voting turnout ~ates for different groups
of voters. Because individual voting data was used in this study,
we can see voting turnout rates for different ethnic and age voting
groups. The following results show that there were major
differences in voting turnout for different ethnic groups in Miami
Beach. Regardless of ethnicity, there also were strong age
differences in voting rates, with middle-age and older voters
voting at a much higher rate than younger voters. In looking at
turnouts for ethnic-age groups, it is also important to note the
sizes of the three age groups comprising each ethnic subgroup. A
large number of young voters lowered the overall voting rate for
each group, and the large turnout for older Hispanic voters for all
Hispanic subgroups raised the overall turnouts for these subgroups.
Out of a total of 34,777 registered voters, 10,689 voted, for
a total turnout of 30.7%. Hispanic voters accounted for 37.2% of
the registered voters, but because Hispanic voters had a higher
voting turnout, they accounted for 42.6% of all votes cast. Because
of the higher turnout, Hispanic voters accounted for the majority
of the voters in 14 of the 31 precincts. (Precinct 26 no longer
exists, but 14 voters still were listed under that precinct. Thus,
the plots indicate 32 precincts visually, but prec~nct 26 counts
4
Table 1. Voter Turnout by Ethnic and Age Subgroups
Ethnic Subgroups
Age Groupings of Voters
18-40 41-64 65+
Foreign-born Hispanics
(N=9,625)*
15.2
(1781)
Cuban-born
(N=6,553)
18.5
(861)
Non-Cuban Hisp.Foreign-born
(N=3,072)
12.2
(920)
Puerto Rican-born
(N=978)
10.2
(372 )
Colombian-born
(N=541)
10.7
(131)
Hispanic Self-Identity**
(N=1,242)
13 .1
(665)
U,S.-Born Hispanic Sample
(Sample N=319)
(Estimated N=2,077)
11. 2
(251)
34.4
(3861)
39.8
(2427)
25.2
(1434)
22.8
(412)
25.2
(286)
38.3
(313)
27.1
(48)
56.1
(3983 )
59.0
(3265)
42.6
(718)
36.1
(194)
48.4
(124)
60.6
(264 )
55.0
(20 )
Total
39.8
46.6
25.4
20.7
27.0
29.5
16.3
------------------------------------------------------------------
Jewish Sample
(Sample N=970)
(Estimated N=13,142)
12.8
(125)
"Other" Sample***
(Sample N=l,337)
(Estimated N=9,290)
13 .4
(545)
37.1
(213)
28.2
(454)
34.0
(632)
38.2
(338)
32.0
24.7
------------------------------------------------------------------
Total Hispanics
(Estimated N=12,944)
13 .2
(4080)
Total non-Hispanics
(Estimated N=21,833)
15.2
(6178)
34.1
(4487)
29.7
(5961)
56.3
(4377)
35.3
(9690)
35.1
28.1
------------------------------------------------------------------
Total Voters
(N=34,777)*
14 .5
(10258)
31.6
(10448)
41.9
(14067)
30.7
*A few voters are loss due to lack of information.
**Those voters who registered for the first time in 1995, and self-
identified themselves as Hispanic by ethnicity, but did not have a
space in which to list place of birth.
***Non-Hispanic and non-Jewish.
5
very little in the statistical operation because of the size}.
As shown in Table 1, the turnout varied by large amounts for
different ethnic and age subgroups. The three largest subgroups of
Hispanic voters are those from Cuba, Puerto Rico, and Colombia.
Hence, these three groups are shown separately {for foreign-born
only because national-origins heritage for United States-born
Hispanic voters are not known} . Results also are shown for a sample
of Jewish voters, by age groups, using the Distinctive Jewish Names
Technique to obtain a sample. Results also are shown for a sample
of non-Jewish, non-Hispanic voters by using a list of surnames.
These are estimates, based on samples, but generally are basically
accurate as estimates.
Results
Despite the federal judicial opinion regarding ethnically
polarized voting in Miami Beach, this sociological and demographic
analyis indicates statistically that ethnically polarized voting
continues in Miami Beach. All four elections of November 1995 were
characterized by ethnic polarization. The least polarization was in
the mayor's election between Seymour Gelber and Andrew Delaplaine,
in which there was no Hispanic candidate. The correlation between
ethnicity and voting was .56, but the level of significance was
.0008, suggesting that these ethnic differences would be found by
chance only about eight times out of ten thousand.
The most polarized voting was found in the election between
Susan Gottlieb and Ada Llerandi, involving a non-Hispanic incumbent
and a Hispanic challenger. The correlation was .93, and the level
of significance was .0000, suggesting that these differences would
be found by chance less than one time out of ten thousand.
The election between Marty Shapiro and Matti Bower also
involved a non-Hispanic incumbent and a Hispanic challenger. This
race was less ethnically polarized than tbe election between Susan
Gottlieb and Ada Llerandi, but the correlation was .79. While not
as strong as the Gottlieb-Llerandi difference, this difference also
was significant beyond the .0000 level, suggesting that these
ethnic differences would be found by chance less than one time out
of ten thousand.
The vote on whether to change the name of Ocean Beach Park to
Marjorie Stoneman Douglas Park also exhibited strong ethnic
polarization. The correlation between ethnicity and voting was .83,
significant beyond the .0000 level. This suggests that these ethnic
voting patterns also would be found by chance less than one time in
ten thousand.
In summary, all four elections exhibited significant ethnic
voting polarization. The results from least polarized to most
polarized with ethnic registration-voting correlations are:
Seymour Gelber and Andrew Delaplaine .56
Martin Shapiro and Matti Bower .79
Keep Same Name or Change Name of Park .83
Susan Gottlieb and Ada Llerandi .93
This summary is further supported visually by the findings at
6
the end of this report. The results for the two elections with
Hispanic and non-Hispanic candidates are shown numerically in
Appendix A, and graphically in Appendix B. On each graph, each
point represents a precinct. The lines that connect the points show
the percentage of voters who are either Hispanic or "Anglo." These
graphs are not weighted for precinct size, but show how each
candidate's vote varies (by precinct) from the line which shows the
percentages of Hispanic or "Anglo" voters in each precinct. These
graphs, without weighting and without any sophisticated statistical
procedures, simply show how the votes received by these four
candidates varies according to the ethnicity of the precincts. As
seen from the graphs, as the number of Hispanic voters increases,
the number of votes for Ada Llerandi and Matti Bower usually
increase, and as the number of Anglo voters increases, the votes
for Susan Gottlieb and Marty Shapiro usually increase. These graphs
indicate that ethnic polarization did occur, to a degree that is
obvious simply by looking at the graphs.
References
Lavender, Abraham D. "Political Implications of Demographic Changes
in Miami Beach From 1980 to 1990: A Look at Blacks, Hispanics,
Jews, and Others." Florida International University, April 1991, 19
pages.
Lavender, Abraham D. "Changes Among Miami Beach Voters From 1991 to
1993: Comparisons of Ethnicity, Age, Geographical Origins, and
Poli tical Affiliations." Florida International University, May
1993, 23 pages.
Lavender, Abraham D., with the assistance of Chris Girard, Ethnic
Voting in Miami Beach From 1983 Through 1993: Homogeneous Precinct
Analysis and Bivariate Ecological Regresion Analyis as Measures of
Ethnically Polarized Voting. Florida International University,
September 1994, 141 pages. Available from the author, Department of
Sociology and Anthropology, FlU, North Miami, Florida 33181,
telephone 305-919-5859.
Lavender, Abraham D. "The Distinctive Hispanic Names (DHN)
Technique: A Method for Selecting a Sample or Estimating Population
Size." Names: A Quarterly Journal, Volume 40, Number 1, March 1992,
pages 1-16.
Rosenwaike, Ira. "Leading Surnames Among American Jews." Names: A
Quarterly Journal, Volume 38, March-June 1990, pages 31-38.
Appreciation is expressed to Joe Geller, Chair, and Gus Garcia,
Executive Director, of the Dade County Democratic Party, and to
Jose Prendes of FlU's Computer Science Center for their support of
this research project.
8
"p c.i."
024
029
r""". 'C..I'''.'
I
~
,
.
i
.ISLANOOO
~..,
8
Left: All precincts in
Miami Beach
Below: Larger-scale of
precincts in South
Beach
1995 MAYOR
BIVARIATE ECOLOGICAL REGRESSION RESULTS
SEYMOUR GELBER AND ANDREW DELAPLAINE
TOTAL NUMBER OF REGISTERED VOTERS
NUMBER OF HISPANIC VOTERS
PERCENT OF TOTAL VOTERS WHO ARE HISPANIC (FOREIGN & U.S.)
PERCENT OF TOTAL VOTES CAST BY HISPANIC VOTERS:
34,777
12,945
37.2
42.6
TOTAL VOTES COUNTED
9,966
Candidates:
Number of Votes
% of Votes
Seymour Gelber
Andrew Delaplaine
7,090
2,876
71.1
28.9
TOTAL VOTES COUNTED
9,966
100.0
RESULTS OF BIVARIATE ECOLOGICAL REGRESSION ANALYSIS
% of Non-Hispanic
Votes Received
% of Hispanic
Votes Received
Seymour Gelber
Andrew Delaplaine
83.6
16.4
54.7
45.3
Total
100.0
100.0
This election had two non-Hispanic candidates, but there was
a distinct difference in voting results for Hispanic voters and
non-Hispanic voters. The correlation coefficient between Hispanic
voters and Seymour Gelber was -.56, while the correlation between
Hispanic voters and Andrew Delaplaine was +.56. These correlation
coefficients were significant at the .0008 level, indicating that
these differences would be found by chance about eight times out of
ten thousand.
9
PLOT OF ELECTION FOR MAYOR BETWEEN
ANDREW DELAPLAINE AND SEYMOUR GELBER
Based on Bivariate Ecological Regression Analysis.
--Each Square Represents a Precinct
--Vertical axis (up and down) = percent of votes received by
Andrew Delaplaine (first candidate alphabetically)
--Horizontal axis (left to right) = percent of voters who are
Hispanic
--Conclusion: As the percentage of Hispanic voters increased, the
percentage of votes obtained by Andrew Delaplaine increased
~
n
~
.
---
~
'.
.
.
.
---
o
o
~
~
n
~
10
1995 MAYOR HOMOGENEOUS PRECINCT ANALYSIS RESULTS
SEYMOUR GELBER AND ANDREW DELAPLAINE
MOST "ANGLO" PRECINCTS
Precincts included in analysis: 23, 24, 25, 27, 28, 30
OF TOTAL VOTES CAST, PERCENT CAST BY "ANGLOS":
OF TOTAL VOTES CAST, PERCENT CAST BY HISPANICS:
81.3%
18.7%
TOTAL VOTES COUNTED
2,469
Candidates:
Number of Votes
Seymour Gelber
Andrew Delaplaine
1,951
518
J. 'r~ ?
% of Votes
1:1
79 .~
21. 0
COMPARISON OF ECOLOGICAL REGRESSION ANALYSIS AND HOMOGENEOUS
PRECINCT ANALYSIS (USING MOST "ANGLO" PRECINCTS):
Of the 2,469 votes counted in this cluster of precincts, 18.7%
(462) were cast by Hispanics and 81.3% (2,007) were cast by non-
Hispanics. The regression analysis suggested that Seymour Gelber
received 54.7% of the Hispanic votes (462 X 54.7% = 253 votes) and
83.6% of the non-Hispanic votes (2,007 X 83.6% = 1,678 votes). The
Hispanic 253 and the non-Hispanic 1,678 total 1,931. The actual
number of votes received was 1,951, suggesting that the estimate
obtained by the regression analysis is very close to the actual
number received. . Because there were only two candidates, the
numbers for Andrew Delaplaine are the reverse of these and hence
are not computed.
MOST HISPANIC PRECINCTS (FOR INFORMATION ONLY, NO ANALYSIS)
Precincts included: 13, 32, 34, 41, 43, 44, 46, 48
OF TOTAL VOTES CAST, PERCENT CAST BY "ANGLOS": 37.4%
OF TOTAL VOTES CAST, PERCENT CAST BY HISPANICS: 62.6%
TOTAL VOTES COUNTED
2,366
Candidates:
Number of Votes
% of Votes
Seymour Gelber
Andrew Delaplaine
1,598
768
67.5
32.5
11
1995 COMMISSION GROUP V
BIVARIATE ECOLOGICAL REGRESSION RESULTS
SUSAN GOTTLIEB AND ADA LLERANDI
TOTAL NUMBER OF REGISTERED VOTERS
NUMBER OF HISPANIC VOTERS
PERCENT OF TOTAL VOTERS WHO ARE HISPANIC (FOREIGN & U.S.)
PERCENT OF TOTAL VOTES CAST BY HISPANIC VOTERS:
34,777
12,945
37.2
42.6
TOTAL VOTES COUNTED
10,076
Candidates:
Number of Votes
% of Votes
Susan Gottlieb
Ada Llerandi
5,457
4,619
54.2
45.8
TOTAL VOTES COUNTED
10,076
100.0
RESULTS OF BIVARIATE ECOLOGICAL REGRESSION ANALYSIS
% of Non-Hispanic
Votes Received
% of Hispanic
Votes Received
Susan Gottlieb
Ada Llerandi
90.8
9.2
2.6
97.4
Total
100.0
100.0
This election had one non-Hispanic candidate and one Hispanic
candidate, and there was a very strong ethnically polarized pattern
when comparing the difference in voting results for Hispanic voters
and non-Hispanic voters. The correlation coefficient between
Hispanic voters and Susan Gottlieb was -.93, while the correlation
between Hispanic voters and Ada Llerandi was +.93. These
correlation coefficients were significant beyond the .0000 level,
indicating that these differences would be found by chance less
than one time out of ten thousand.
12
PLOT OF ELECTION FOR COMMISSIONER BETWEEN
SUSAN GOTTLIEB AND ADA LLERANDI
Based on Bivariate Ecological Regression Analysis
--Each Square Represents a Precinct
--Vertical axis (up and down) = percent of votes received by Susan
Gottlieb (first candidate alphabetically)
- -Horizontal axis (left to right) = percent of voters who are
Hispanic
--Conclusion: As the percentage of Hispanic voters increased, the
percentage of votes obtained by Susan Gottlieb decreased
100
75
>0
2S
1$
...
13
1995 COMMISSION GROUP V
HOMOGENEOUS PRECINCT ANALYSIS RESULTS
SUSAN GOTTLIEB AND ADA LLERANDI
MOST "ANGLO" PRECINCTS
Precincts included in analysis: 23, 24, 25, 27, 28, 30
OF TOTAL VOTES CAST, PERCENT CAST BY "ANGLOS": 81.3%
OF TOTAL VOTES CAST, PERCENT CAST BY HISPANICS: 18.7%
TOTAL VOTES COUNTED 2,489
Candidates:
Number of Votes
% of Votes
Susan Gottlieb
Ada Llerandi
1,854
635
74.5
25.5
COMPARISON OF ECOLOGICAL REGRESSION ANALYSIS AND HOMOGENEOUS
PRECINCT ANALYSIS (USING MOST "ANGLO" PRECINCTS):
Of the 2,489 votes counted in this cluster of precincts, 18.7%
(465) were cast by Hispanics and 81.3% (2,023) were cast by non-
Hispanics. The regression analysis suggested that Susan Gottlieb
received 2.6% of the Hispanic votes (465 X 2.6% = 12 votes) and
90.8% of the non-Hispanic votes (2,023 X 90.8% = 1,837 votes). The
Hispanic 12 and the non-Hispanic 1,837 total 1,849, The actual
number of votes received was 1,854, suggesting that the estimate
obtained by the regression analysis is very close to the actual
number received. Because there were ,only two candidates, the
numbers for Ada Llerandi are the reverse of these and hence are not
computed.
MOST HISPANIC PRECINCTS (FOR INFORMATION ONLY, NO ANALYSIS)
Precincts included: 13, 32, 34, 41, 43, 44, 46, 48
OF TOTAL VOTES CAST, PERCENT CAST BY "ANGLOS": 37.4%
OF TOTAL VOTES CAST, PERCENT CAST BY HISPANICS: 62.6%
TOTAL VOTES COUNTED
2,378
Candidates:
Number of Votes
% of Votes
Susan Gottlieb
Ada LLerandi
841
1,537
35.4
64.6
14
1995 COMMISSION GROUP VI
BIVARIATE ECOLOGICAL REGRESSION RESULTS
MARTY SHAPIRO AND MATTI BOWER
TOTAL NUMBER OF REGISTERED VOTERS
NUMBER OF HISPANIC VOTERS
PERCENT OF TOTAL VOTERS WHO ARE HISPANIC (FOREIGN & U.S.)
PERCENT OF TOTAL VOTES CAST BY HISPANIC VOTERS:
34,777
12, 945
37.2
42.6
TOTAL VOTES COUNTED
9,983
Candidates:
Number of Votes
% of Votes
Martin Shapiro
Matti Bower
5,225
4,758
52.3
47.7
TOTAL VOTES COUNTED
9,983
100.0
RESULTS OF BIVARIATE ECOLOGICAL REGRESSION ANALYSIS
% of Non-Hispanic
Votes Received
% of Hispanic
Votes Received
Martin Shapiro
Matti Bower
74.2
25.8
22.8
77.2
Total
100.0
100.0
This election had one non-Hispanic candidate and one Hispanic
candidate, and there was a fairly strong ethnically polarized
pattern when comparing the difference in voting results for
Hispanic voters and non-Hispanic voters. The correlation
coefficient between Hispanic voters and Martin Shapiro was -.79,
while the correlation between Hispanic voters and Matti Bower was
+.79. These correlation coefficients were significant beyond the
.0000 level, indicating that these differences would be found by
chance less than one time out of ten thousand.
15
PLOT OF ELECTION FOR COMMISSIONER BETWEEN
MARTY SHAPIRO AND MATTI BOWER
Based on Bivariate Ecological Regression Analysis
--Each Square Represents a Precinct
- -Vertical axis (up and down) = percent of votes received by
Matti Bower (first candidate alphabetically)
--Horizontal axis (left to right)
Hispanic
percent of voters who are
--Conclusion: As the percentage of Hispanic voters increased, the
percentage of votes obtained by Matti Bower increased
.,.
50
75
25
o
o
25
50
75
100
16
1995 COMMISSION GROUP VI
HOMOGENEOUS PRECINCT ANALYSIS RESULTS
MARTIN SHAPIRO AND MATTI BOWER
MOST "ANGLO" PRECINCTS
Precincts included in analysis: 23, 24, 25, 27, 28, 30
OF TOTAL VOTES CAST, PERCENT CAST BY "ANGLOS": 81.3%
OF TOTAL VOTES CAST, PERCENT CAST BY HISPANICS: 18.7%
TOTAL VOTES COUNTED 2,477
Candidates:
Number of Votes
% of Votes
Martin Shapiro
Matti Bower
1,601
876
64.6
35.4
COMPARISON OF ECOLOGICAL REGRESSION ANALYSIS AND HOMOGENEOUS
PRECINCT ANALYSIS (USING MOST "ANGLO" PRECINCTS):
Of the 2,477 votes counted in this cluster of precincts, 18.7%
(463) were cast by Hispanics and 81.3% (2,014) were cast by non-
Hispanics. The regression analysis suggested that Marty Shapiro
received 22.8% of the Hispanic votes (463 X 22.8% = 106 votes) and
74.2% of the non-Hispanic votes (2,014 X 74.2% = 1,494 votes). The
Hispanic 106 and the non-Hispanic 1,494 totals 1,600. The actual
number of votes received was 1,601, suggesting that the estimate
obtained by the regression analysis is very close to the actual
number received. Because, there were only two candidates, the
numbers for Matti Bower are the reverse of these and hence are not
computed.
MOST HISPANIC PRECINCTS (FOR INFORMATION ONLY, NO ANALYSIS)
Precincts included: 13, 32, 34, 41, 43, 44, 46, 48
OF TOTAL VOTES CAST, PERCENT CAST BY "ANGLOS": 37.4%
OF TOTAL VOTES CAST, PERCENT CAST BY HISPANICS: 62.6%
TOTAL VOTES COUNTED
2,348
Candidates:
Number of Votes
% of Votes
Martin Shapiro
Matti Bower
979
1,369
41.7
58.3
17
1995 PARK RENAMING
BIVARIATE ECOLOGICAL REGRESSION RESULTS
KEEP SAME NAME OR CHANGE NAME
TOTAL NUMBER OF REGISTERED VOTERS
NUMBER OF HISPANIC VOTERS
PERCENT OF TOTAL VOTERS WHO ARE HISPANIC (FOREIGN & U.S.)
PERCENT OF TOTAL VOTES CAST BY HISPANIC VOTERS:
34,777
12,945
37.2
42.6
TOTAL VOTES COUNTED
8,642
Ballot Choices:
Number of Votes
% of Votes
No, Keep Name as Ocean Beach Park
Yes, Change Name to M.S. Douglas
4,392
4,250
50.8
49.2
TOTAL VOTES COUNTED
8,642
100.0
RESULTS OF BIVARIATE ECOLOGICAL REGRESSION ANALYSIS
% of Non-Hispanic
Votes Received
% of Hispanic
Votes Received
No, Keep Same Name
Yes, Change Name
37.5
62.5
69.3
30.7
Total
100.0
100.0
This election was on an issue with no direct reference to
ethnicity, but there was a fairly strong ethnically polarized
pattern when comparing the difference in voting results for
Hispanic voters and non-Hispanic voters. The correlation
coefficient between Hispanic voters and favoring a name change to
Majorie Stoneman Douglas was -.83, while the correlation between
Hispanic voters and opposition to the name change was +.83. These
correlation coefficients were significant beyond the .0000 level,
indicating that these differences would be found by chance less
than one time out of ten thousand.
18
PLOT OF ELECTION ON WHETHER
TO KEEP NAME AS OCEAN BEACH PARK
OR CHANGE NAME TO MARJORIE STONEMAN DOUGLAS PARK
Based on Bivariate Ecological Regression Analysis
--Each Square Represents a Precinct
--Vertical axis (up and down) = percent of votes received by
"Yes, Change Name" (first choice on ballot)
--Horizontal axis (left to right)
Hispanic
percent of voters who are
--Conclusion: As the percentage of Hispanic voters increased, the
percentage of "Yes, Change Name" votes decreased
100
"
..,
. .
.
. .
.
.
.
..
.
.
.
-
25
o
o
Z5
50
75
IlO
19
1995 PARK RENAMING
HOMOGENEOUS PRECINCT ANALYSIS RESULTS
KEEP SAME NAME OR CHANGE NAME
MOST "ANGLO" PRECINCTS
Precincts included in analysis: 23, 24, 25, 27, 28, 30
OF TOTAL VOTES CAST, PERCENT CAST BY "ANGLOS": 81.3%
OF TOTAL VOTES CAST, PERCENT CAST BY HISPANICS: 18.7%
TOTAL VOTES COUNTED 2,140
Ballot Choices:
Number of Votes
% of Votes
No, Keep Name as Ocean Beach Park
Yes, Change Name to M.S. Douglas
949
1,191
44.3
55.7
COMPARISON OF ECOLOGICAL REGRESSION ANALYSIS AND HOMOGENEOUS
PRECINCT ANALYSIS (USING MOST "ANGLO" PRECINCTS):
Of the 2,140 votes counted in this cluster of precincts, 18.7%
(400) were cast by Hispanics and 81.3% (1,740) were cast by non-
Hispanics. The regression analysis suggested that "Yes, Change
Name" received 30.7% of the Hispanic votes (400 X 30.7% = 123
votes) and 62.5% of the non-Hispanic votes (1,740 X 62.5% = 1,087
votes). The Hispanic 123 and the non-Hispanic 1,087 totals 1,210.
The actual number of votes received was 1,191, suggesting that the
estimate obtained by the regression analysis is very close to the
actual number received. Because there were only two choices, the
numbers for "No, Keep Same Name" are the reverse of these and hence
are not computed.
MOST HISPANIC PRECINCTS (FOR INFORMATION ONLY, NO ANALYSIS)
Precincts included: 13, 32, 34, 41, 43, 44, 46, 48
OF TOTAL VOTES CAST, PERCENT CAST BY "ANGLOS": 37.4%
OF TOTAL VOTES CAST, PERCENT CAST BY HISPANICS: 62.6%
TOTAL VOTES COUNTED
1,954
Ballot Choices: Number of Votes
% of Votes
No, Keep Name as Ocean Beach Park 1,132
Yes, Change Name to M.S. Douglas 822
57.9
42.1
20
APPENDIX A
HISPANIC AND NON-HISPANIC PERCENTAGE OF VOTES IN EACH PRECINCT
AND VOTES FOR HISPANIC AND NON-HISPANIC CANDIDATES
Column 2 gives the percentage of total votes in each precinct which
were cast by Hispanic voters. Columns 3 and 4 give the percentage
of votes received by Ada Llerandi and Matti Bower. Column 5 gives
the percentage of total votes in each precinct which were cast by
"Anglo" (Non-Hispanic) voters. Columns 6 and 7 give the percentage
of votes received by Susan Gottlieb and Marty Shapiro. In general,
as the percentage of Hispanic voters increases, the percentage of
votes received by Ada Llerandi and Matti Bower increases, and as
the percentage of non-Hispanic voters increases, the percentage of
votes received by Susan Gottlieb and Marty Shapiro increases.
Pcnt.
#
28
25
30
24
27
23
38
31
29
15
14
37
40
11
19
39
42
18
21
35
22
33
36
46
41
32
13
48
43
44
34
Absent
Total
% Hisp.
Votes
13.9
15.3
17.8
19.5
21. 0
24.2
27.2
27.7
30.2
31.4
35.1
38.6
44.0
46.4
47.8
47.8
48.5
50.6
52.9
53.7
54.5
54.7
58.9
59.1
60.3
60.9
64.3
65.7
65.9
66.5
68.7
42.6
% Votes
Llerandi
26.7
26.7
23.7
25.2
33.3
23.6
32.0
37.9
35.8
25.9
45.3
41.1
49.1
39.7
40.1
52.3
60.8
49.8
53.1
63.6
51.1
60.2
61.2
70.6
62.1
52.4
60.1
74.5
68.4
72.9
73.1
42.8
45.8
% Votes
Bower
32.4
26.0
43.2
45.9
38.5
24.2
51. 7
50.4
41.3
38.1
38.0
43.2
46.9
36.8
50.0
42.7
59.9
49.3
48.1
55.6
55.6
56.6
64.7
60.8
61.4
48.4
56.2
64.3
58.5
60.4
67.5
52.4
47.7
21
% N-H
Votes
% Votes
Gottlieb
86.1
84.7
82.2
80.5
79.0
75.8
72.8
72 .3
69.8
68.6
64.9
61.4
56.0
53.6
52.2
52.2
51. 5
49.4
47.1
46.3
45.5
45.3
41.1
40.9
39.7
39.1
35.7
34.3
34.1
33.5
31.3
73.3
73.3
76.3
74.8
66.7
76.4
68.0
62.1
64.2
74.1
54.7
58.9
50.9
60.3
59.9
47.7
39.2
50.2
46.9
36.4
48.9
39.8
38.8
29.4
37.9
47.6
39.9
25.5
31.6
27.1
26.9
57.2
54.2
57.4
% Votes
Shapiro
67.6
74.0
56.8
54.1
61.5
75.8
48.3
49.6
58.7
61. 9
62.0
56.8
53.1
63.2
50.0
57.3
40.1
50.7
51. 9
44.4
44.4
43.4
35.3
39.2
38.6
51. 6
43.8
35.7
41.5
39.6
32.5
47.6
52.3
APPENDIX Bl
GRAPHIC REPRESENTATION OF RELATIONSHIP, BY PRECINCTS, BETWEEN
PERCENTAGE OF HISPANIC VOTERS AND VOTES FOR ADA LLERANDI.
Small squares represent percentage of Hispanic voters for each
precinct, and plus signs represent votes for Ada Llerandi.
80
40
60
20
J
I I
, I
GRAPHIC REPRESENTATION OF RELATIONSHIP, BY PRECINCTS, BETWEEN
PERCENTAGE OF HISPANIC VOTERS AND VOTES FOR MATTI BOWER.
Small squares represent percentage of Hispanic voters for each
precinct, and asterisks represent votes for Matti Bower.
eo
70
10
60
60
30
20
o
22
APPENDIX B2
GRAPHIC REPRESENTATION OF RELATIONSHIP, BY PRECINCTS, BETWEEN
PERCENTAGE OF "ANGLO" VOTERS AND VOTES FOR SUSAN GOTTLIEB.
Diamonds represent percentage of "Anglo" voters for each precinct,
and large squares represent votes for Susan Gottlieb.
100
so
60
40 r
I
I
20
o
GRAPHIC REPRESENTATION OF RELATIONSHIP, BY PRECINCTS, BETWEEN
PERCENTAGE OF "ANGLO" VOTERS AND VOTES FOR MARTY SHAPIRO.
Diamonds represent percentage of "Anglo" voters for each precinct,
and x's represent votes for Marty Shapiro.
100
so
60
40
20
o
23
APPENDIX B3
GRAPHIC REPRESENTATION OF RELATIONSHIP, BY PRECINCTS; BETWEEN
PERCENTAGB OF HISPANIC VOTERS, VOTES FOR ADA LLERANDI, AND VOTES
FOR MATTI BOWER.
Small squares represent percentage of Hispanic voters for
precinct, plus signs represent votes for Ada Llerandi,
asterisks represent votes for Matti Bower.
each
and
801
!
,
l
60 ~
t[! ,,'
, ,
40 ~*
t: , /\,
I
01
r I
GRAPHIC REPRESENTATION OF RELATIONSHIP, BY PRECINCTS, BETWEEN
PERCENTAGE OF "ANGLO" VOTERS, VOTES FOR SUSAN GOTTLIEB, AND VOTES
FOR MARTY SHAPIRO.
Diamonds represent percentage of "Anglo" voters for each precinct,
large squares represent votes for Susan Gottlieb, and x's represent
votes for Marty Shapiro.
1001
SO
60
40
20
0 I I
24
.
"
.
APPENDIX B4
GRAPHIC REPRESENTATION OF RELATIONSHIP, BY PRECINCTS, BETWEEN
PERCENTAGE OF "ANGLO" VOTERS AND VOTES FOR ADA LLERANDI, AND
PERCENTAGE OF "ANGLO" VOTERS AND VOTES FOR MATTI BOWER.
Diamonds represent percentage of "Anglo" voters, plus signs
represent votes for Ada Llerandi, and asterisks represent votes for
Matti Bower.
100
60
~ //\}
80
40
100
60
so
20
o
25
f
,
.
.
APPENDIX BS
GRAPHIC REPRESENTATION OF RELATIONSHIP, BY PRECINCTS, BETWEEN
PERCENTAGE OF HISPANIC VOTERS AND VOTES FOR SUSAN GOTTLIEB, AND
PERCENTAGE OF HISPANIC VOTERS AND VOTES FOR MARTY SHAPIRO.
Small squares represent percentage of Hispanic voters, large
squares represent votes for Susan Gottlieb, and x's represent votes
for Marty Shapiro.
80
60
40
20
o
80
60
o
26